Machine Learning and Knowledge Extraction

@EricTopol @AdeAdamson @NEJM Indeed the fact that "just a consensus among pathologists doesn't solve the problem" is a critical but oft-neglected issue in ML. w\ @cabitzaf we recently discussed this, as related to explainable AI and reliability of "Gold"

@RogueRad @woojinrad @DrLukeOR @EricTopol How to pass from multi-rater labelling to "sound" ground truth, or how to make input explainable, not only #AI: our proposal is out at last: https://t.co/QcRN3otGM1